Lithospheric Structure in the Southwestern United States
The southwestern United States hosts a number of interesting features:
Contrasts in topography and surface deformation between the Basin and Range Province, the Colorado Plateau, and the Rio Grande Rift.
Widespread volcanism, despite the location far from the plate boundary.
Strong variations in seismic velocity throughout the lithosphere (learn more about this in my Tomography page).
The goal of this research is to understand variations in lithospheric properties, including thickness and seismic wave speed structures. What controls surface deformation and topography? What is the role of melting within or just below the lithosphere?
The high topography of the Colorado Plateau has puzzled geophysicists for decades. It is also responsible for spectacular erosional features such as the Grand Canyon, as shown in a photo I took from the North Rim in September 2021.
First, we need to understand the first-order structure: how thick the lithosphere is, and how this relates to the various tectonic units. To do this, I use a technique called receiver functions, which identifies where major interfaces are located based on where seismic waves are converted. If seismic velocity changes rapidly, for instance at the boundary between the lithosphere and asthenosphere, an S wave will convert part of its energy into a P-wave. Receiver functions show us the depth where these conversions happen, and a further step called Common Conversion Point stacking, or CCP stacking, connects different receiver functions from different locations, and allows us to map out these boundaries in three dimensions. From this, I derived a map of the inferred Lithosphere-Asthenosphere Boundary (LAB).
a) Depth of LAB below the southwestern US. The diamonds represent locations of volcanics, colored according to age. b) Location of shallower gradient than LAB (inferred Mid-Lithospheric Discontinuity), with volcanic ages of 1 million years or younger only. Names of geologic provinces: BNR - Basin and Range; CP - Colorado Plateau; RG - Rio Grande Rift; SN - Sierra Nevada Batholith. Figure from Golos & Fischer (2022).
These maps shed new light on some of the variations between geologic regions in the southwestern US. The lithosphere is thicker below the Colorado Plateau than the Basin and Range Province; the amplitude of the CCP stack is also stronger below the Basin and Range. Partially molten rock just below the lithosphere is one explanation for strong CCP stack signals in the LAB. This would also explain why the most recent volcanoes are most likely to occur above lithosphere that is 80 km thick, in agreement with geochemical analysis of samples from volcanoes.
For more information on this study, check out our recent paper here!
The Next Step:
Finding the depth of the LAB is only one part of this study. I also want to quantify the bulk Vs in order to investigate the differences in seismic velocity within the lithosphere. This will allow us to understand how intralithospheric structures and wave speed anomalies might be related to chemical and thermal processes within the Earth.
I am jointly inverting information from the CCP stacks alongside surface wave information. The CCP stacks have information about where velocity gradients (i.e. those big boundaries in Earth’s structure) occur, and how strong those gradients are, but are not sensitive to absolute wavespeed. Meanwhile, surface waves do constrain absolute wave speed, so combining these two types of data will yield a more complete understanding of shear velocity structure at lithospheric depths, including a detailed view of the LAB.
To combine surface waves and body waves, unlike my tomography work which is a linear problem, I am using a Bayesian inversion. This is a probabilistic method: I generate many test models, and evaluate which characteristics give the best fit to data. The advantages of this method are that 1) it allows users to combine different types of information that might not necessarily follow the same mathematical relationship, and 2) by finding a group of models with acceptable fits, we can quantify the uncertainty in the solution.
This is a work in progress. More information and results to come soon!